CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices...

76
CHAPTER SIX CHAPTER SIX Eigenvalues Eigenvalues

Transcript of CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices...

Page 1: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

CHAPTER SIXCHAPTER SIX

EigenvaluesEigenvalues

Page 2: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Outlines

System of linear ODE (Omit)DiagonalizationHermitian matricesOuadratic formPositive definite matrices

Page 3: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

MotivationsTo simplify a linear dynamics such

that it is as simple as possible.To realize a linear system characteristics

e.g.,

the behavior of system dynamics.

xAx

zecharacteripofroot

bapeqch

byyay

)(

0)(..

0

2

Page 4: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Example: In a town, each year 30% married women get divorced 20% single women get married In the 1st year, 8000 married women 2000 single

women. Total population remains constant

be the women numbers at year i,

where represent married & single women

respectively.

si

mi

iW

WWLet

si

mi WW &

ii WWW

8.03.0

2.07.0&

2000

800010

Page 5: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

If

Question: Why does converges?

Why does it converges to the same limit even

when the initial condition is different?

12,&6000

4000,,

4000

6000

2000

8000,

8.03.0

2.07.0

12121

01

nWWWW

WWAWW

n

iii

14,

6000

4000

0

10000

14

14

0

iWW

W

W

i

iW

Page 6: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Ans: Choose a basis

Given an initial for some

for example

Question: How does one know choosing such a basis?

1

1&

3

221 xx

2211 2

1& xxAxxA

1

1)4000(

3

22000

2000

8000

221100 xCxCWW 21 &CC

nasxCxACxACWA nnn1122110

Page 7: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Def: Let . A scalar is said to be an

eigenvalue or characteristic value of A if

such that

The vector is said to be an eigenvector

or characteristic vector belonging to .

is called an eigen pair of A.

Question: Given A, How to compute eigenvalues

& eigenvectors?

xxA

nnFA

0x

x

),( x

Page 8: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

is an eigen pair of

is singular

Note that, is a polynomial, called

characteristic polynomial of A, of degree n in

Thus, by FTA, A has exactly n eigenvalues including

multiplicities. is a eigenvector associated with

eigenvalue while is eigenspace of A.

0)det(

),(

0,0)(

0,

IA

IA

xxIA

xxxA

nnFA

.

)det()( AIp

0\)( AINx

),( x

)( IAN

Page 9: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Example: Let

are eigenvalues of A. To find eigenspace of 2:( i.e., )

3&2

)3)(2(65

11

24det

)det()(

11

24

2

AIp

A

1

1)2(

0

0

11

220)2(

2

1

spanIAN

x

xxIA

)2( IAN

Page 10: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

To find eigenspaces of 3(i.e., )

Let

1

2)3(

0

0

21

210)3(

2

1

spanIAN

x

xxIA

)3( IAN

30

02

11

21

1App

p

Page 11: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Let . Then

is an eigenvalue of A.

has a nontrivial solution.

is singular.

loses rank.

nIANullityvii

IAvi

IAv

IAiv

IANiii

xIAii

i

)()(

)(

0)det()(

)(

0)()(

0))((

)(

nnFA

Page 12: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Let .

If is an eigenvalue of A with eigenvector

Then

This means that is also an eigen-pair of

A.

xxxAxA

xxA

nnA

.xC

),( x

Page 13: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Let .

Where are eigenvalues of A.

(i) Let

(ii) Compare with the coefficient of , we

have

nnA

)()det()(1

i

n

iAIp

n 1

n

iiA

1

det0

1n)(

11

Atracean

iii

n

ii

Page 14: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Theorem 6.1.1: Let

Then and consequently A & B have the

same eigenvalues.

Pf: Let for some nonsingular matrix S.

)()det(

)det()det()det(

)(det

)det(

)det()(

1

1

1

A

B

PAI

SAIS

SAIS

ASSI

BIP

BA ~

)()( BA PP

ASSB 1

Page 15: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Diagonalization

Goal: Given find nonsingular matrix S

a diagonal matrix.

Question1: Are all matrices diagonalizable?

Question2: What kinds of A are diagonalizable?

Question3: How to find S if A is diagonalizable?

nnFA

DASS 1

Page 16: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

NOT all matrices are diagonalizable

e.g., Let

If A is diagonalizable

nonsingular matrix S

00

10A

2

11

0

0

d

dASS

00

0

0

0)det(

0)(

1

2

1

21

21

21

Sd

dSA

dd

ddA

ddAtrace

Page 17: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

To answer Q2,

Suppose A is diagonalizable.

nonsingular matrix S

Let

are eigenpair of A for

This gives a condition for and diagonalizability and a way

to find S.

nSSS 1

ni ,,1

SDAS

)( 11

ndddiagDASS

nSASA 1 nn SdSd 11

ii Sd

Page 18: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Theorem 6.3.2: Let is diagonalizable

A has n linear independent eigenvectors.

Note : Similarity transformation

Change of coordinate

diagonalization

)()( EE

WL

A

B

1 SASB

)()( EE

WL

SI

nnFA

xPy

SI

Page 19: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Theorem6.3.1: If are distinct eigenvalues of a

matrix A with corresponding eigenvectors

, then are linear independent.

Pf: Suppose are linear independent

not all zero

Suppose

are distinct.

are linear independent.

.... golw

0)(2

iii

K

ixCIA

n 1

0)(2

11

K

ii xC

nn

01

i

K

ii xC

Kxx 1 Kxx 1

Kxx 1

KCC 1

01 C

K

i

xx

C

x

1

1

1

*0

&0

Page 20: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Remarks: Let , and

(i) is an eigenpair of A for

(ii) The diagonalizing matrix S is not unique because

Its columns can be reordered or multiplied by

an nonzero scalar

(iii) If A has n distinct eigenvalues , A is diagonalizable.

If the eigenvalues are not distinct , then may or may not

diagonalizable depending on whether or not A has n

linear independent eigenvectors.

(iv)

1SDSA )( 1 ndiagD nSSS 1

ii S ni ,,1

11

1 SdiagSSSDA nkk

Page 21: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Example: Let

For

For

Let

112

202

213

A

100

010

0001SDS

1,1,00det AI

1

1

1

)0(,0 SpanIAN

1

2

0

,

0

2

1

)(,1 SpanIAN

101

221

011

S

Page 22: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Def: If an matrix A has fewer than nlinear independent eigenvectors,we say that A is defective

e.g.

(i) is defective

(ii) is defective

10

11A

nn

00

10A

Page 23: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Example: Let A & B both have the same eigenvalues Nullity (A-2I)=1 The eigenspace associated with has only one dimension. A is NOT diagonalizable However, Nullity (B-2I)=2 B is diagonalizable

263

041

002

&

201

040

002

BA

4,2,2

2

001

020

000

)2(

rankIArank

1

063

021

000

)2(

rankIBrank

2

Page 24: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Question:

Is the following matrix diagonalizable ?

00

00

01

)(

00

10

01

)( BiiAi

Page 25: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

The Exponential of a Matrix Motiration :

Motiration:The general solution of

is

The unique solution of

is

Question:What is and

how to compute ?

xAx cetx At)(

00 )( xtx

xAx

0)( 0)( xetx ttA

AteAte

Page 26: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Note that

Define

0 !i

ia

i

ae

......!2!

2

0

AAI

i

Ae

i

iA

Page 27: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Suppose A is diagonalizable with

1

00

)(!

SDS

i

Ae

i

i

i

iAt

1SDSA

kSSDA kk ,1

1 SSeDt

1

0

01

S

e

e

St

t

n

Page 28: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Example: Compute

Sol: The eigenvalues A are

with eigenvectors

0,1

1

3

1

221 xandx

00

01

11

32;1 DXXDXA

11

10

0

X

eXXXee DA

231

6623

ee

ee

Page 29: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Hermitian matrices : Let , then A can

be written as

where

e.g. ,

ii

iiA

274

32

21

11

74

32i

nnCA

iCBA

nnRCB ,

Page 30: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Let , then

e.g. ,

ii

ii

ii

iiH

273

42

274

32

nnCA TH AA

HHH

HHH

HH

ACAC

BABA

AA

Page 31: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Def:

A matrix A is said to be Hermitian if

A is said to be skew-Hermitian if

A is said to be unitary if

( → its column vectors form an orthonormal set in )

HAA

AAH

IUU H nC

Page 32: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Theorem6.4.1: Let Then

(i)

(ii) eigenvectors belonging to distinct eigenvalues are orthogonal

Pf : (i)Let be an eigenpair of A

(ii) Let and be two eigenpairs of A with

HAA RA

x,) õjjxjj2= xHAx = xHAHx = (xHAx)H = õöjjxjj2

) õ = õö ) õ 2 R(õ1;x1) (õ2;x2)

õ16=õ2) (õ1à õ2)xH1x2= õ1xH1x2à õ2xH1x2

= (õ1x1)Hx2à xH1 (õ2x2)

= xH1AHx2à xH1Ax2= 0

) xH1x2= 0) x1 ? x2

(i)

Page 33: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Theorem: Let Then

Pf : (i) Let be an eigenpair of A

is pure-imaginary

HAA

axisjA

x,

xxxAx

xAxxxxxHH

HHHH

AAH

Page 34: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Theorem6.4.3: (Schur`s Theorem) Let

Then unitary matrix U is

upper triangular

Pf : Let be an eigenpair of A with

Choose to be such that is unitary

Choose

Chose to be unitary

Continue this process , we have the theorem

nnCA AUU H

11,w 11 w

nww 2 nwwU 11

2

11

1

1

11

*0

*

nTn

T

H wAwA

w

w

AUU

222

211

2112

2

2

*0

*

0

01

ww

w

UAUUU

w

UH

HH

2w

)1()1(4

32

222

0

nn

Hww

0

0

0

2

1

2112

UUAUU HH

Page 35: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Theorem6.4.4: (Spectral Theorem)

If , then unitary matrix U

that diagonalizes A .

Pf : By previous Theorem , unitary matrix

, where

T is upper triangular .

T is a diagonal matrix

AAH

TAUUU H

TAUU

UAUAUUTH

HHHHH

AAH

Page 36: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Cor: Let A be real symmetric matrix .Then (i) (ii) an orthogonal matrix U is a diagonal matrixRemark : If A is Hermitian , then , by Th6.4.4 , Complete orthonormal eigenbasis

RA

AUU T

Page 37: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Example:

Find an orthogonal matrix U that diagonalizes A

Sol : (i)

(ii)

(iii)By Gram-Schmidt Process

The columns of

form an orthogonormal eigenbasis (WHY?)

021

232

120

ALet

TT

T

SpanAIN

SpanAIN

0,1,2,1,0,1)(

6

1,

6

2,

6

1)5(

)5()1()det()( 2 AIp5,1,1

TT

SpanAIN3

1,

3

1,

3

1,

2

1,0,

2

1)(

3

1

2

1

6

13

10

6

23

1

2

1

6

1

U

1,1,5 diagAUU T

Page 38: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Question:In addition to Hermitian matrices ,

Is there any other matrices possessing orthonormal

eigenbasis?

Note : If A has orthonormal eigenbasis

where U is Hermitian &

D is diagonal

HUDUA

HHHH

HHHH

HHHH

AAUUDUDU

UUDDDUUD

UDUUUDAA

Page 39: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Def: A is said to be normal if

Remark : Hermitian , Skew- Hermitian and Unitary matrices are all normal

HH AAAA

Page 40: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Theorem6.4.6: A is normal A possesses orthonormal eigenbasisPf : have proved By Th.6.4.3 , unitary U is upper triangular T is also normal Compare the diagonal elements of

T has orthonormal eigenbasis(WHY?)

""""

AUUT HHH AAAA

HH TTTT

HH TTTT &

2

1

2

2

2

2

2

1

2

2

1

2

1

2

11

nn

n

iin

n

jj

ii

n

jj

tt

tt

tt

jiTij ,0

Page 41: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Singular Value Decomposition(SVD) :

Theorem : Let with rank(A)=r

Then unitary matrices

With

Where

nmCA nmCU

rrr

nn RdiagCV 1&021 r

Hi

r

iii

H

nm

rrH

uu

VVUUVUA

1

212100

0

nrr

mrr

vvVvvV

uuUuuU

1211

1211

,

,

Page 42: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Remark:In the SVD

The scalars are called

singular values of A

Columns of U are called

left singular vectors of A

Columns of V are called

right singular vectors of A

HVUA

r 1

Page 43: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Pf : Note that , is Hermitian

& Positive semidefinite with

unitary matrix V

where

Define

(1)

(2)

Define (3)

Define is unitary

nnH CAA

rAArank H )(

H

rH VdiagVAA 00,22

1 021 r

)(211 )()(& rnnrnr VVVdiag

HH VVAA 1

2

1

00 22 AVAVAH

rmCAVU 1

11

)3)(1(r

H IUU 11

00

0,HVUA

00

02121 UUVVA

mmCUUU 212 )3)(2(

Page 44: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Remark:In the SVD The singular values of A are unique while U&V are not unique Columns of U are orthonormal eigenbasis for Columns of V are orthonormal eigenbasis for

is an orthonormal basis for

is an orthonormal basis for

is an orthonormal basis for

is an orthonormal basis for

HVUA r 1

AAHHAA

VUA

UUAV

H

00

0

nr vv 1

mr uu 1

ruu 1

rvv 1

)( HAN

)( HAR

)(AN

)(AR

Page 45: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

rank(A) = number of nonzero singular values

but rank(A) ≠ number of nonzero eigenvalues

for example

0,0)(1)(

00

10

AbutArankA

Page 46: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Example : Find SVD of

Sol :

An orthonormal eigenbasis associate with

can be found as

Find U is orthogonal

A set of candidate for are

Thus

00

11

11

A

0,40,4)( 21 AAT

AAH

11

11

2

1V

TvAu )0,2

1,

2

1(

11

11

)()(, 32 VUAANuu TT

32 &uu

3311

2 ,)0,2

1,

2

1(

1euvAu T

00

00

02

withVUA T

Page 47: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Lemma6.5.2 : Let

be orthogonal . Then

Pf :

mmnm RQRA &

FFAQA

2

1

2

2

1

2

2

21

2

F

n

ii

n

ii

FnF

A

A

QA

QAQAQA

Page 48: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Cor : Let be

the SVD of A . Then

nnH RVUA

n

iiFF

A1

2

Page 49: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

We`ll state the next result without proof :

Theorem6.5.3 :

H.(1) be the SVD of A

(2)

C :

nnH RVUA

kSrankRS nm )(

FS

n

kji

F

k

i

Tiii ASvuA

min1

2

1

Page 50: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Application : Digital Image Processing

p. 377

(especially efficient for matrix which

has low rank)

Page 51: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Quadratic Forms :

To classify the type of quadratic surface (line) Optimization : An application to the Calculus

Page 52: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Def : A quadratic equation in two variables x & y

is an equation of the form

02 22 feydxcybxyax

0),(),(

f

y

xed

y

x

dc

bayx

Page 53: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Standard forms of conic sections

(i)

(ii)

(iii)

(iv)

Note : Is there any difference between the eigenvalues

A of the quadratic form ?XAX T

circlery

xyxryx

2222

10

01)(

ellipsey

x

b

ayxb

y

a

x

11

0

01

)(1

2

2

2

2

2

2

hyperbolay

x

b

ayxb

y

a

x

11

0

01

)(1

2

2

2

2

2

2

0010

00)(22

y

x

y

xyxyxorxy

Page 54: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Goal : Try to transform the quadratic equation

into standard form by suitable translation

and rotation

Page 55: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Example : (No xy term)

The eigenvalues of the quadratic terms are 9 , 4

→ ellipse

011164189 22 yyxx

362419 22 yx

1

3

2

2

12

2

2

2

yx

)21(

Page 56: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Example : (Have xy term)

→ By direct computation

is orthogonal

( why does such U exist ? )

→ Let the original equation becomes

08323 22 yxyx

831

13)(

y

xyx

00

00

45cos45sin

45sin45cos

2

1

2

12

1

2

1

,40

02UUUA T

y

xU

y

x

840

02)(

y

xyx

x

y

x

y

045

Page 57: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Example :

→ Let

or

or

0428323 22 yyxyx

04)280(31

13)(

y

x

y

xyx

2

1

2

12

1

2

1

, Uy

xU

y

x

x

y

x

y

4)280(40

02)(

y

xU

y

xyx

48)(4)(2 22 yxyx

8)1(2)2( 22 yx

1&2,

8)(2)( 22

yyxx

yx

x

y

Page 58: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Optimization :

Let

It is known from Taylor’s Theorem of Calculus that

Where is Hessian matrix

is local extremum

If

then is a local minimum

2: CRRf n

...

)()())(()()( 00000

TOH

xxHxxxxxfxfxf T

nnxx RxfHji

))(( 0

))(,( 00 xfx 0)( 0 xf

)0.(

0)()(0&0)( 000

resp

xxHxxxf T

}0{)( 00* xxxxNx

))(,( 00 xfx .)max.(resp

Page 59: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Def : A real symmetric matrix A is said to be

(i) Positive definite denoted by

(ii) Negative definite denoted by

(iii) Positive semidefinite denoted by

(iv) Negative semidefinite denoted by

example :

is indefinite

question : Given a real symmetric matrix , how to determine

its definiteness efficiently ?

0,00 xAxxifA T

0,00 xAxxifA T

0,00 xAxxifA T

0,00 xAxxifA T

040

02

A

040

02

A

040

02

A

Page 60: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Theorem6.5.1: Let Then

Pf : let be eigenpair of A

Suppose

Let be an orthonormal eigen-basis of A

(Why can assume this ? )

""

""

nnT RAA RAA )(0

),(&0 xA 2

xxAxT

02 x

xAxT

RA)(}{ 1 nxx

0A

n

n

iii

n someforxXRx 11

,

)?(0

)()(

1

2

11

why

xxxAx

n

iii

n

iiii

Tn

iii

T

Page 61: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Example: Find local extrema of

Sol :

Thus f has local maximum at while

are saddle points

22 cosy3)( zxzxzyxf

)0,0,0()2siny,3,2( zxzxf

)0,,0()y,,( nzx

201

030

102

)0,2,0( nH

1,3,3)( H

201

030

102

)0,)12(,0( nH

1,3,3)( H)0,2,0( n

)0,)12(,0( n

Page 62: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Positive Definite Matrices :

Property I : P is nonsingular

Property II :

and all the leading principal submatrices of

A are positive definite

0& PRPP nnT

0det(P)& )det(P)( i

iPPRPP iinnT ,00&

Page 63: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Property III :

P can be reduced to upper triangular form

using only row operation III and the pivots

elements will all be positive

Sketch of the proof :

& determinant is invariant under row

operation of type III

Continue this process , the property can be proved

0& PRPP nnT

011 P

22

1211

2221

12112 0 P

PP

PP

PPP

02 P0)1(

22 P

Page 64: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Property IV : Let Then

(i) A can be decompose as A=LU where L is lower triangular & U is upper triangular

(ii) A can be decompose as A=LU where L is lower triangular & U is upper triangular with all the

diagonal element being equal to 1 , D is an

diagonal matrix

Pf : by Gaussian elimination and the fact that the

product of two lower (upper) triangular matrix

is lower (upper) triangular

0& ARAA nnT

Page 65: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Example:

Thus A=LU

Also A=LDU with

522

2102

224

A

430

390

2242

1

2

11213

AEE

UAE

300

390

2243

123

13

1

2

1

012

1000

3

1

2

1

2

1 where 231312 EEEL

1003

110

2

1

2

11

&

300

090

004

,

13

1

2

1

012

1001

UDL

Page 66: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Property V : Let

If

Pf :

LHS is lower triangular & RHS is upper triangular

with diagonal elements 1

0& ARAA nnT

212121

222111

&,

then,

UUDDLL

UDLUDLA

11211

12

12

UUDLLD

121

12 UUIUU

IDLLD 11

12

12

(why?) 11

21

1211

2 ILLDDLL

211

1221 DDIDDLL

Page 67: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Property VI : Let

Then A can be factored into

where D is a diagonal matrix & L is lower triangular

with 1’s along the diagonal

Pf :

Since the LDU representation is unique

0& ARAA nnT

TTAA DLULDU

LDUAT

Let

TLDLA

TLU

Page 68: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Property VII : (Cholesky decomposition)

Let

Then A can be factored into where

L is lower triangular with positive diagonal

Hint :

TT LDLDLDLA ))(( 2

1

2

1

0& ARAA nnT

TLLA

Page 69: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Example: We have seen that

Note that

Define we have the

Choleoky decomposition

LUA

300

390

224

13

1

2

1

012

1001

522

2102

224

LDUA

1003

110

2

1

2

11

300

090

004

13

1

2

1

012

1001

Also

TTT LDLALUAA

LDL 2

1

1

TLLA 11

Page 70: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Theorem6.6.1 : Let , Then the followings are equivalent:

(i) A>0

(ii) All the leading principal submatrices have positive determinants.

(iii) A ~ U only using elementary row operation of type III. And the pivots are all positive , where U is an upper triangular matrix.

(iv) A has Cholesky decomposition LLT.

(v) A can be factored into BTB for some nonsingular matrix B

nnT RAA

row

Pf : We have shown that (i) (ii) (iii) (iv) In addition , (iv) (v) (i) is trivial

Page 71: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Housholder Transformation :

Def : Then the matrix

is called

Housholder transformation

Geometrical lnterpretation:

Q is symmetric ,

Q is orthogonal ,

What is the eigenvalues , eigenvectors

and determinant of Q ?

1Let 2u

TuuIQ 2

QQT 1, QQIQQ TT

Page 72: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Given , Find

x u

1)2( exuuIxH T

2221 xxHe

1

1

ex

exu

QR factorization

xu

1e

....

ˆ0

01,

*0

*0

*...

*0

*......

222

21

HHHH

HHAHH

k

kk

Page 73: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Theorem : Let and be a

SVD of A with

Then

Pf :

Cor : Let be nonsingular with

singular values

Then and

nnRA TVUA )( 1 ndiag

)(max12AAA T

12

max

12

max

1

2

max

12

max

12

yxVyxV

xVUxAA

y

TT

x

T

xx

nnRA

n 1

ii 0n

A1

2

1

Page 74: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Application : In solving What is the effect of the

Solution when present measurement error ?

bxA b

bxAbb~~ and ~

~Let

)1(~

.~ 1 bbAxx

b

bbAA

x

xx

A

bx

xAxAb

~

..~

)2(

. , Also

1)2)(1(

Page 75: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

is said to be the condition number of A

If A is orthogonal then

This means that , due to the error in b

the deviation of the associated solution of

is minimum if A is orthogonal

1.)( AAAK

nn

n

RAAK , 1)( 1

1)( AK

bAx

Page 76: CHAPTER SIX Eigenvalues. Outlines System of linear ODE (Omit) Diagonalization Hermitian matrices Ouadratic form Positive definite matrices.

Example :

A is close to singular

Note that , is the solution for and is the solution for

What does this mean ?

Similarly , i.e. small deviation in x results in large deviation in b This is the reason why we use orthogonal factorization in Numerical solving Ax=b

11

001.11A

1.)(& AAAK

1

1x

001.2

001.2b

0

2~x

2

2~b

1~

but22

001.0~

x

xx

b

bb

1

0~ ,

0

1

1001

1001~ , 1000

1000bbxx